NetSuite expands Text Enhance feature

NetSuite, a business solutions provider, announced during its SuiteConnect event in New York City that its generative AI-driven Text Enhance feature, launched last year, can now be applied to 200 additional places across its entire product line.

The feature is, essentially, an AI-driven text generator and editor that lets users draw on company-specific and relevant data from NetSuite to access product content such as item descriptions, sales communication, purchase orders, job listings, employee goals, notes and other long-form text fields throughout the system.

When TextEnhance launched last year, it was available in all modules, but could only be used on certain text boxes. he latest version further expands the feature across the entire NetSuite oeuvre through the addition of over 200 "use cases," or new text boxes where the AI can be deployed.

"Today we're expanding Text Enhance to all areas of the suite to automatically generate professional writing in over 200 new fields in over 100 experiences," said Evan Goldberg, founder and executive vice president of Oracle NetSuite. "Your team can save time on notes, email production, often with better consistency with your brand's tone and style with fewer errors,"

In terms of finance and accounting, this means the feature can now be applied to journal entries to help describe transactions and new accounts, as well as to purchase order entries such as packing list and product labeling instructions. The feature can also now be used to write cash refund explanations to maintain consistency for customer communications and help internal auditors. This is in addition to the existing use cases of summarizing narratives for financial reports and creating personalized collection letters.

There are similar product enhancements for supply chain and operations, manufacturing, sales and marketing, human resources and customer support.

Brian Chess, NetSuite's AI development leader, said the large language model underpinning Text Enhance is built on the Oracle Cloud Infrastructure, which provides generative AI service to customers by, among other things, offering trained models for various purposes. Chess said Text Enhance specifically uses the model from Cohere LLM, a company which recently partnered with NetSuite.

The model itself is housed within NetSuite's own servers, as opposed to connecting to another company's, according to Chess. This means no customer data is shared with large language model providers or seen by other customers. In addition, an individual customer is the only entity allowed to use custom models trained on its data. Furthermore, role-based security is embedded directly into NetSuite workflows and only recommends content that end users are entitled to view.

Goldberg, in a later press conference, referenced Oracle's history as a security-minded company to further address concerns about exposure of user data to an LLM. He noted that the team behind the OCI infrastructure is very concerned about this exact same problem and added that Oracle itself has spent its over 40 years of existence well aware of the importance of security.

"[We have] 40-plus years of experience we can bring to analyze how these LLMs are using your data and make sure that data is secure," he said. "We put [data] in specific places, and we control, sort of, the prompts that go to the system. We're an intermediary there, so that is another way to control what gets sent out to the world."

This specificity of data placement was also why he was confident that LLMs' propensity to give information made up out of whole cloth was less of an issue. Essentially, the LLM is drawing on a very limited set of what he called "indisputable facts" about the company using it, meaning it has less room to get creatively inaccurate.

"We're not asking the system to opine on if war is necessary. I think just the relatively limited domain we have can help with accuracy," he said, adding, "We expect humans to read it. We're not going full self-driving quite yet."

Chess, the AI lead, said that having the model on Oracle servers is a cost saver because companies often charge for API calls, which can add up over time. By keeping its LLM in-house, NetSuite avoids incurring such charges. Goldberg said he does not plan to implement similar charges for NetSuite customers in the immediate future, and if there ever are charges for generative AI services, it will only be for very resource-intensive applications.

"We could charge by use of AI so every time you click Text Enhance it costs three cents, but we're not going that route and I don't think we ever will," he said.

Goldberg, during the press conference, shared some of his wider views on AI and NetSuite's plans. Overall, Goldberg believes AI will become ubiquitous in society, to the point where buying software without it will eventually be like buying a car without wheels. He noted that this is part and parcel of the rising demand for tech proficiency in younger generations that motivated the company's investment in AI.

"It pushed us and drove us to double down and build the next-generation business experience and that experience has to include AI. It needs to be everywhere, embedded in the system, not something you choose to use, but just the natural output," he said.

To illustrate, he raised the example of an AI that notices people with similar profiles as the user. This tool might say, "We've noticed other users with a similar profile like to see this information [on their dashboard] and maybe this is something you'd want to do. Totally right there, proactive in the user experience as a part of it, not a thing along the side, but always there whenever you ask for something in the system," he said.

For reprint and licensing requests for this article, click here.
Technology Artificial intelligence Machine learning Hardware and software
MORE FROM ACCOUNTING TODAY